Guest Editors’ Introduction
Cloud Computing
Distributed Internet Computing for IT
and Scientific Research
O
Marios D. Dikaiakos ne vision of 21st century com-
and George Pallis puting is that users will access
University of Cyprus Internet services over lightweight
portable devices rather than through
Dimitrios Katsaros some descendant of the traditional
University of Thessaly desktop PC. Because users won’t have
(or be interested in) powerful machines,
Pankaj Mehra who will supply the computing power?
Hewlett-Packard Labs The answer to this question lies with
cloud computing.
Athena Vakali Cloud computing is a recent trend
Aristotle University of Thessaloniki in IT that moves computing and data
away from desktop and portable PCs
into large data centers. It refers to
applications delivered as services over
the Internet as well as to the actual
cloud infrastructure — namely, the
hardware and systems software in data
centers that provide these services.
The key driving forces behind cloud
computing are the ubiquity of broad-
band and wireless networking, falling
storage costs, and progressive improve-
ments in Internet computing software.
Cloud-service clients will be able to
add more capacity at peak demand,
reduce costs, experiment with new ser-
10 Published by the IEEE Computer Society 1089-7801/09/$26.00 © 2009 IEEE
vices, and remove unneeded capacity,
whereas service providers will increase
utilization via multiplexing, and allow
for larger investments in software and
hardware.
Currently, the main technical under-
pinnings of cloud computing infrastruc-
tures and services include virtualization,
service-oriented software, grid comput-
ing technologies, management of large
facilities, and power efficiency. Consum-
ers purchase such services in the form
of infrastructure-as-a-service (IaaS),
platform-as-a-service (PaaS), or soft-
ware-as-a-service (SaaS) and sell value-
added services (such as utility services)
to users. Within the cloud, the laws of
probability give service providers great
leverage through statistical multiplexing
of varying workloads and easier man-
agement — a single software installation
can cover many users’ needs.
We can distinguish two differ-
ent architectural models for clouds:
the first one is designed to scale out
by providing additional computing
instances on demand. Clouds can use
IEEE INTERNET COMPUTING

these instances to supply services in the form
of SaaS and PaaS. The second architectural
model is designed to provide data and compute-­
intensive applications via scaling capacity. In
most cases, clouds provide on-demand comput-
ing instances or capacities with a “pay-as-you-
go” economic model. The cloud infrastructure
can support any computing model compatible
with loosely coupled CPU clusters. Organiza-
tions can provide hardware for clouds internally
(internal clouds), or a third party can provide
it externally (hosted clouds). A cloud might be
restricted to a single organization or group (pri-
vate clouds), available to the general public over
the Internet (public clouds), or shared by mul-
tiple groups or organizations (hybrid clouds).
A cloud comprises processing, network, and
storage elements, and cloud architecture con-
sists of three abstract layers. Infrastructure is
the lowest layer and is a means of delivering
basic storage and compute capabilities as stan-
dardized services over the network. Servers,
storage systems, switches, routers, and other
systems handle specific types of workloads,
from batch processing to server or storage
augmentation during peak loads. The middle
platform layer provides higher abstractions
and services to develop, test, deploy, host, and
maintain applications in the same integrated
development environment. The application
layer is the highest layer and features a com-
plete application offered as a service.
Key Challenges
In 1961, John McCarthy envisioned that “com-
putation may someday be organized as a public
utility.” We can view the cloud computing para-
digm as a big step toward this dream. To realize
it fully, however, we must address several sig-
nificant problems and unexploited opportunities
concerning the deployment, efficient operation,
and use of cloud computing infrastructures.
Software/Hardware Architecture
Cloud computing services’s emergence suggests
fundamental changes in software and hardware
architecture. Computer architectures should
shift the focus of Moore’s law from increasing
clock speed per chip to increasing the number
of processor cores and threads per chip. Indus-
try and academia must design novel systems
and services that would exploit a high degree
of parallelism. Software architectures for mas-
SEPTEMBER/OCTOBER 2009
Guest Editors’ Introduction
sively parallel, data-intensive computing, such
as MapReduce (http://labs.google.com/papers/
mapreduce.html), will grow in popularity. In
terms of storage technologies, we’ll likely shift
from hard disk drives (HDDs) to solid-state
drives (SDDs), such as flash memories, or, given
that completely replacing hard disks is pro-
hibitively expensive, hybrid hard disks — that
is, hard disks augmented with flash memories,
which provide reliable and high-performance
data storage. The biggest barriers to adopting
SSDs in data centers have been price, capacity,
and, to some extent, the lack of sophisticated
query-processing techniques. However, this
is about to change as SSDs’ I/O operations per
second (IOPS) benefits become too impressive to
ignore, their capacity increases at a fast pace,
and we devise new algorithms and data struc-
tures tailored to them.
Data Management
The shift of computer processing, storage, and
software delivery away from desktop and local
servers, across the Internet, and into next-­
generation data centers results in limitations
as well as new opportunities regarding data
management. Data is replicated across large
geographic distances, where its availability and
durability are paramount for cloud service pro-
viders. It’s also stored at untrusted hosts, which
creates enormous risks for data privacy. Com-
puting power in clouds must be elastic to face
changing conditions. For instance, providers
can allocate additional computational resources
on the fly to handle increased demand.
They should deploy novel data management
approaches, such as analytical data manage-
ment tasks, multitenant databases for SaaS, or
hybrid designs among database management
systems (DBMSs) and MapReduce-like systems
so as to address data limitations and harness
cloud computing platforms’ capabilities.
Cloud Interoperability
Cloud interoperability refers to customers’ abil-
ity to use the same artifacts, such as manage-
ment tools, virtual server images, and so on,
with a variety of cloud computing providers
and platforms.
Cloud interoperability will enable cloud
infrastructures to evolve into a worldwide,
transparent platform in which applications aren’t
restricted to enterprise clouds and cloud service
11
Cloud Computing
providers. We must build new standards and
interfaces that will enable enhanced portability
and flexibility of virtualized applications. Up to
now, significant discussion has occurred around
open standards for cloud computing. In this con-
text, the “Open Cloud Manifesto” (www.open
cloudmanifesto.org) provides a minimal set of
principles that will form a basis for initial agree-
ments as the cloud community develops stan-
dards for this new computing paradigm.
Security and Privacy
In cloud computing, a data center holds infor-
mation that end-users would more tradition-
ally have stored on their computers. This
raises concerns regarding user privacy protec-
tion because users must outsource their data.
Additionally, the move to centralized services
We must build new standards and
interfaces that will enable enhanced
portability and flexibility of virtualized
applications.
could affect the privacy and security of users’
interactions. Security threats might happen
in resource provisioning and during distrib-
uted application execution. Also, new threats
are likely to emerge. For instance, hackers can
use the virtualized infrastructure as a launch-
ing pad for new attacks. Cloud services should
preserve data integrity and user privacy. At
the same time, they should enhance interoper-
ability across multiple cloud service providers.
In this context, we must investigate new data-
protection mechanisms to secure data privacy,
resource security, and content copyrights.
Service Provisioning and Cloud Economics
Providers supply cloud services by signing
service-level agreements (SLAs) with consum-
ers and end-users. Cloud service consumers, for
instance, might have an SLA with a cloud ser-
vice provider concerning how much bandwidth,
CPU, and memory the consumer can use at any
given time throughout the day. Underestimat-
ing the provision of resources would lead to
broken SLAs and penalties. On the other hand,
12 www.computer.org/internet/
overestimating the provision of resources would
lead to resource underutilization and, conse-
quently, a decrease in revenue for the provider.
Deploying an autonomous system to efficiently
provision services in a cloud infrastructure is a
challenging problem due to the unpredictability
of consumer demand, software and hardware
failures, heterogeneity of services, power man-
agement, and conflicting signed SLAs between
consumers and service providers.
In terms of cloud economics, the provider
should offer resource-economic services. Novel,
power-efficient schemes for caching, query
processing, and thermal management are man-
datory due to the increasing amount of waste
heat that data centers dissipate for Internet-
based application services. Moreover, new pric-
ing models based on the pay-as-you-go policy
are necessary to address the highly variable
demand for cloud resources.
In this Issue
Given the continued, intense activity in the
cloud arena, we invited researchers and practi-
tioners to submit articles to this special issue of
IC describing research efforts and experiences
concerning the deployment, efficient operation,
and use of cloud computing infrastructures.
From among the 42 submissions, and after rig-
orous review, we selected the following four
articles as representative of ongoing research
and development activities.
The first article, “Virtual Infrastructure
Management in Private and Hybrid Clouds,”
by Borja Sotomayor, Rubén S. Montero, Igna-
cio M. Llorente, and Ian Foster, presents two
open source projects for private and hybrid
clouds. OpenNebula is a virtual infrastructure
manager that can be used to deploy virtualized
services on both a local pool of resources and
on external IaaS clouds. Haizea is a resource
lease manager that can act as a scheduling
back end for OpenNebula, providing advance
reservations and resource preemption.
“Harnessing Cloud Technologies for a Virtu-
alized Distributed Computing Infrastructure,” by
Alexandre di Costanzo, Marcos Dias de Assun-
ção, and Rajkumar Buyya, presents the realiza-
tion of a system — termed the InterGrid — for
interconnecting distributed computing infra-
structures by harnessing virtual machines. The
article provides an abstract view of the proposed
architecture and its implementation. Experi-
IEEE INTERNET COMPUTING

ments show the scalability of an InterGrid-
managed infrastructure and how the system can
benefit from using cloud infrastructure.
In “Content-Centered Collaboration Spaces
in the Cloud,” John S. Erickson, Susan Spence,
Michael Rhodes, David Banks, James Ruther-
ford, Edwin Simpson, Guillaume Belrose, and
Russell Perry envision a cloud-based platform
that inverts the traditional application-content
relationship by placing content rather than
applications at the center, letting users rapidly
build customized solutions around their content
items. The authors review the dominant trends
in computing that motivate the exploration of
new approaches for content-centered collabo-
ration and offer insights into how certain core
problems for users and organizations are being
addressed today.
The final article, “Sky Computing,” by
Katarzyna Keahey, Maurício Tsugawa, Andréa
Matsunaga, and José A.B. Fortes, describes
the creation of environments configured on
resources provisioned across multiple distrib-
uted IaaS clouds. This technology is called
sky computing. The authors provide a real-
world example and illustrate its benefits with
a deployment in three distinct clouds of a bio­
informatics application.
C loud computing is a disruptive technol-
ogy with profound implications not only
for Internet services but also for the IT sector
as a whole. Its emergence promises to stream-
line the on-demand provisioning of software,
hardware, and data as a service, achieving
economies of scale in IT solutions’ deployment
and operation. This issue’s articles tackle top-
ics including architecture and management
of cloud computing infrastructures, SaaS and
IaaS applications, discovery of services and
data in cloud computing infrastructures, and
cross-platform interoperability.
Still, several outstanding issues exist, par-
ticularly related to SLAs, security and pri-
vacy, and power efficiency. Other open issues
include ownership, data transfer bottlenecks,
performance unpredictability, reliability, and
software licensing issues. Finally, hosted appli-
cations’ business models must show a clear
pathway to monetizing cloud computing. Sev-
eral companies have already built Internet
consumer services such as search, social net-
SEPTEMBER/OCTOBER 2009
Guest Editors’ Introduction
working, Web email, and online commerce that
use cloud computing infrastructure. Above all,
cloud computing’s still unknown “killer appli-
cation” will determine many of the challenges
and the solutions we must develop to make this
technology work in practice.
Acknowledgments
We would like to express our gratitude to the authors of
all submitted articles and the reviewers for their contri-
butions to this special issue. We thank Fred Douglis, IC’s
editor in chief, and Michael Rabinovich, associate editor
in chief, for their support of the special issue, and also the
production staff at the IEEE Computer Society who made
this issue possible.
Marios D. Dikaiakos is an associate professor at the Univer-
sity of Cyprus, Nicosia. His research interests include
network-centric computing, with an emphasis on grids,
vehicular computing, and Web technologies. Dikaiakos
has a PhD in computer science from Princeton Univer-
sity. He’s a senior member of the ACM, and a member
of the IEEE Computer Society and the Technical Cham-
ber of Greece. Contact him at mdd@cs.ucy.ac.cy.
Dimitrios Katsaros is a lecturer at the University of Thes-
saly, Greece. His research interests include distributed
systems, such as the Web and Internet, social networks,
mobile and pervasive computing, and wireless ad hoc
and wireless sensor networks. Katsaros has a PhD in
informatics from Aristotle University of Thessaloniki.
Contact him at dkatsar@inf.uth.gr.
Pankaj Mehra is a principal member of the technical staff at
Hewlett-Packard Labs. His research focuses on the design
of systems and networks for large-scale enterprise appli-
cations. Contact him at pankaj.mehra@hp.com.
George Pallis is a visiting lecturer at the University of
Cyprus, Nicosia. His research interests include dis-
tributed systems, such as the Web and grids, content
distribution networks, information retrieval, and data
clustering. Pallis has a PhD in informatics from Aristo-
tle University of Thessaloniki. Contact him at ­g pallis@
cs.ucy.ac.cy.
Athena Vakali is an associate professor at the Aristotle Uni-
versity of Thessaloniki. Her current research interests
include Web usage mining, content delivery networks,
Web and social Web data clustering, and Web data
caching/outsourcing. Vakali has a PhD in informatics
from the Aristotle University. Contact her at avakali@
csd.auth.gr.
13